Appropriate use of nomograms to guide prostate cancer treatment selection

For decades physicians have attempted to accurately predict post-treatment outcomes before performing prostate cancer interventions. Use of basic clinical factors, such as clinical T-stage, biopsy Gleason sum, and pretreatment prostate specific antigen, has allowed some level of prediction of pathologic and clinical outcomes. However, these basic tables and risk stratification schema provide a broad range of potential outcomes. The rapid growth of retrospective research in prostate cancer has yielded an abundance of additional potential prognostic factors that may influence outcomes of interest; however, incorporating and understanding the significance of these ever-expanding factors is difficult for even the most experienced physicians. Nomograms incorporate these factors (including treatment-specific) and assign them relative weights to provide a probability of the outcome of interest on a graphical scale. They distill large numbers of data into a manageable format and provide the probability of outcomes on a continuous scale rather than in categoric groups. However, because they require a computation to generate a probability, they are not amenable to memorization, which decreases ease of use. Furthermore, these numbers still have associated confidence intervals and the models are largely derived from retrospective data, which have inherent drawbacks. Clinicians and patients should still exercise due diligence when interpreting the results of these nomograms, and these prediction tools should not serve as a stand-alone substitute for clinical decision-making.